X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=datapath%2Fflow.c;h=d07337c83109c30428b1fc24dcd7917288127c0d;hb=e0edde6fee279cdbbf3c179f5f50adaf0c7c7f1e;hp=27038ecee9854957f1f94df5e35c87373c0974ba;hpb=28bad4735cf1fda0e2064a3128577f3a58d88e8a;p=sliver-openvswitch.git diff --git a/datapath/flow.c b/datapath/flow.c index 27038ecee..d07337c83 100644 --- a/datapath/flow.c +++ b/datapath/flow.c @@ -1,14 +1,24 @@ /* - * Distributed under the terms of the GNU GPL version 2. - * Copyright (c) 2007, 2008, 2009, 2010, 2011 Nicira Networks. + * Copyright (c) 2007-2011 Nicira, Inc. * - * Significant portions of this file may be copied from parts of the Linux - * kernel, by Linus Torvalds and others. + * This program is free software; you can redistribute it and/or + * modify it under the terms of version 2 of the GNU General Public + * License as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA + * 02110-1301, USA */ #include "flow.h" #include "datapath.h" -#include +#include #include #include #include @@ -22,14 +32,13 @@ #include #include #include -#include #include #include #include #include #include #include -#include +#include #include #include #include @@ -37,7 +46,6 @@ #include "vlan.h" static struct kmem_cache *flow_cache; -static unsigned int hash_seed __read_mostly; static int check_header(struct sk_buff *skb, int len) { @@ -48,13 +56,13 @@ static int check_header(struct sk_buff *skb, int len) return 0; } -static inline bool arphdr_ok(struct sk_buff *skb) +static bool arphdr_ok(struct sk_buff *skb) { return pskb_may_pull(skb, skb_network_offset(skb) + sizeof(struct arp_eth_header)); } -static inline int check_iphdr(struct sk_buff *skb) +static int check_iphdr(struct sk_buff *skb) { unsigned int nh_ofs = skb_network_offset(skb); unsigned int ip_len; @@ -73,7 +81,7 @@ static inline int check_iphdr(struct sk_buff *skb) return 0; } -static inline bool tcphdr_ok(struct sk_buff *skb) +static bool tcphdr_ok(struct sk_buff *skb) { int th_ofs = skb_transport_offset(skb); int tcp_len; @@ -89,19 +97,19 @@ static inline bool tcphdr_ok(struct sk_buff *skb) return true; } -static inline bool udphdr_ok(struct sk_buff *skb) +static bool udphdr_ok(struct sk_buff *skb) { return pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct udphdr)); } -static inline bool icmphdr_ok(struct sk_buff *skb) +static bool icmphdr_ok(struct sk_buff *skb) { return pskb_may_pull(skb, skb_transport_offset(skb) + sizeof(struct icmphdr)); } -u64 flow_used_time(unsigned long flow_jiffies) +u64 ovs_flow_used_time(unsigned long flow_jiffies) { struct timespec cur_ts; u64 cur_ms, idle_ms; @@ -115,8 +123,8 @@ u64 flow_used_time(unsigned long flow_jiffies) } #define SW_FLOW_KEY_OFFSET(field) \ - offsetof(struct sw_flow_key, field) + \ - FIELD_SIZEOF(struct sw_flow_key, field) + (offsetof(struct sw_flow_key, field) + \ + FIELD_SIZEOF(struct sw_flow_key, field)) static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key, int *key_lenp) @@ -126,9 +134,10 @@ static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key, int payload_ofs; struct ipv6hdr *nh; uint8_t nexthdr; + __be16 frag_off; int err; - *key_lenp = SW_FLOW_KEY_OFFSET(ipv6.addr); + *key_lenp = SW_FLOW_KEY_OFFSET(ipv6.label); err = check_header(skb, nh_ofs + sizeof(*nh)); if (unlikely(err)) @@ -139,14 +148,23 @@ static int parse_ipv6hdr(struct sk_buff *skb, struct sw_flow_key *key, payload_ofs = (u8 *)(nh + 1) - skb->data; key->ip.proto = NEXTHDR_NONE; - key->ip.tos = ipv6_get_dsfield(nh) & ~INET_ECN_MASK; - ipv6_addr_copy(&key->ipv6.addr.src, &nh->saddr); - ipv6_addr_copy(&key->ipv6.addr.dst, &nh->daddr); + key->ip.tos = ipv6_get_dsfield(nh); + key->ip.ttl = nh->hop_limit; + key->ipv6.label = *(__be32 *)nh & htonl(IPV6_FLOWINFO_FLOWLABEL); + key->ipv6.addr.src = nh->saddr; + key->ipv6.addr.dst = nh->daddr; - payload_ofs = ipv6_skip_exthdr(skb, payload_ofs, &nexthdr); + payload_ofs = ipv6_skip_exthdr(skb, payload_ofs, &nexthdr, &frag_off); if (unlikely(payload_ofs < 0)) return -EINVAL; + if (frag_off) { + if (frag_off & htons(~0x7)) + key->ip.frag = OVS_FRAG_TYPE_LATER; + else + key->ip.frag = OVS_FRAG_TYPE_FIRST; + } + nh_len = payload_ofs - nh_ofs; skb_set_transport_header(skb, nh_ofs + nh_len); key->ip.proto = nexthdr; @@ -162,33 +180,32 @@ static bool icmp6hdr_ok(struct sk_buff *skb) #define TCP_FLAGS_OFFSET 13 #define TCP_FLAG_MASK 0x3f -void flow_used(struct sw_flow *flow, struct sk_buff *skb) +void ovs_flow_used(struct sw_flow *flow, struct sk_buff *skb) { u8 tcp_flags = 0; - if (flow->key.eth.type == htons(ETH_P_IP) && - flow->key.ip.proto == IPPROTO_TCP) { + if ((flow->key.eth.type == htons(ETH_P_IP) || + flow->key.eth.type == htons(ETH_P_IPV6)) && + flow->key.ip.proto == IPPROTO_TCP && + likely(skb->len >= skb_transport_offset(skb) + sizeof(struct tcphdr))) { u8 *tcp = (u8 *)tcp_hdr(skb); tcp_flags = *(tcp + TCP_FLAGS_OFFSET) & TCP_FLAG_MASK; } - spin_lock_bh(&flow->lock); + spin_lock(&flow->lock); flow->used = jiffies; flow->packet_count++; flow->byte_count += skb->len; flow->tcp_flags |= tcp_flags; - spin_unlock_bh(&flow->lock); + spin_unlock(&flow->lock); } -struct sw_flow_actions *flow_actions_alloc(const struct nlattr *actions) +struct sw_flow_actions *ovs_flow_actions_alloc(const struct nlattr *actions) { int actions_len = nla_len(actions); struct sw_flow_actions *sfa; - /* At least DP_MAX_PORTS actions are required to be able to flood a - * packet to every port. Factor of 2 allows for setting VLAN tags, - * etc. */ - if (actions_len > 2 * DP_MAX_PORTS * nla_total_size(4)) + if (actions_len > MAX_ACTIONS_BUFSIZE) return ERR_PTR(-EINVAL); sfa = kmalloc(sizeof(*sfa) + actions_len, GFP_KERNEL); @@ -200,7 +217,7 @@ struct sw_flow_actions *flow_actions_alloc(const struct nlattr *actions) return sfa; } -struct sw_flow *flow_alloc(void) +struct sw_flow *ovs_flow_alloc(void) { struct sw_flow *flow; @@ -216,36 +233,205 @@ struct sw_flow *flow_alloc(void) return flow; } -void flow_free_tbl(struct tbl_node *node) +static struct hlist_head *find_bucket(struct flow_table *table, u32 hash) +{ + hash = jhash_1word(hash, table->hash_seed); + return flex_array_get(table->buckets, + (hash & (table->n_buckets - 1))); +} + +static struct flex_array *alloc_buckets(unsigned int n_buckets) { - struct sw_flow *flow = flow_cast(node); + struct flex_array *buckets; + int i, err; + + buckets = flex_array_alloc(sizeof(struct hlist_head *), + n_buckets, GFP_KERNEL); + if (!buckets) + return NULL; + + err = flex_array_prealloc(buckets, 0, n_buckets, GFP_KERNEL); + if (err) { + flex_array_free(buckets); + return NULL; + } + for (i = 0; i < n_buckets; i++) + INIT_HLIST_HEAD((struct hlist_head *) + flex_array_get(buckets, i)); + + return buckets; +} + +static void free_buckets(struct flex_array *buckets) +{ + flex_array_free(buckets); +} + +struct flow_table *ovs_flow_tbl_alloc(int new_size) +{ + struct flow_table *table = kmalloc(sizeof(*table), GFP_KERNEL); + + if (!table) + return NULL; + + table->buckets = alloc_buckets(new_size); + + if (!table->buckets) { + kfree(table); + return NULL; + } + table->n_buckets = new_size; + table->count = 0; + table->node_ver = 0; + table->keep_flows = false; + get_random_bytes(&table->hash_seed, sizeof(u32)); + + return table; +} + +static void flow_free(struct sw_flow *flow) +{ flow->dead = true; - flow_put(flow); + ovs_flow_put(flow); +} + +void ovs_flow_tbl_destroy(struct flow_table *table) +{ + int i; + + if (!table) + return; + + if (table->keep_flows) + goto skip_flows; + + for (i = 0; i < table->n_buckets; i++) { + struct sw_flow *flow; + struct hlist_head *head = flex_array_get(table->buckets, i); + struct hlist_node *node, *n; + int ver = table->node_ver; + + hlist_for_each_entry_safe(flow, node, n, head, hash_node[ver]) { + hlist_del_rcu(&flow->hash_node[ver]); + flow_free(flow); + } + } + +skip_flows: + free_buckets(table->buckets); + kfree(table); +} + +static void flow_tbl_destroy_rcu_cb(struct rcu_head *rcu) +{ + struct flow_table *table = container_of(rcu, struct flow_table, rcu); + + ovs_flow_tbl_destroy(table); +} + +void ovs_flow_tbl_deferred_destroy(struct flow_table *table) +{ + if (!table) + return; + + call_rcu(&table->rcu, flow_tbl_destroy_rcu_cb); +} + +struct sw_flow *ovs_flow_tbl_next(struct flow_table *table, u32 *bucket, u32 *last) +{ + struct sw_flow *flow; + struct hlist_head *head; + struct hlist_node *n; + int ver; + int i; + + ver = table->node_ver; + while (*bucket < table->n_buckets) { + i = 0; + head = flex_array_get(table->buckets, *bucket); + hlist_for_each_entry_rcu(flow, n, head, hash_node[ver]) { + if (i < *last) { + i++; + continue; + } + *last = i + 1; + return flow; + } + (*bucket)++; + *last = 0; + } + + return NULL; +} + +static void flow_table_copy_flows(struct flow_table *old, struct flow_table *new) +{ + int old_ver; + int i; + + old_ver = old->node_ver; + new->node_ver = !old_ver; + + /* Insert in new table. */ + for (i = 0; i < old->n_buckets; i++) { + struct sw_flow *flow; + struct hlist_head *head; + struct hlist_node *n; + + head = flex_array_get(old->buckets, i); + + hlist_for_each_entry(flow, n, head, hash_node[old_ver]) + ovs_flow_tbl_insert(new, flow); + } + old->keep_flows = true; +} + +static struct flow_table *__flow_tbl_rehash(struct flow_table *table, int n_buckets) +{ + struct flow_table *new_table; + + new_table = ovs_flow_tbl_alloc(n_buckets); + if (!new_table) + return ERR_PTR(-ENOMEM); + + flow_table_copy_flows(table, new_table); + + return new_table; } -/* RCU callback used by flow_deferred_free. */ +struct flow_table *ovs_flow_tbl_rehash(struct flow_table *table) +{ + return __flow_tbl_rehash(table, table->n_buckets); +} + +struct flow_table *ovs_flow_tbl_expand(struct flow_table *table) +{ + return __flow_tbl_rehash(table, table->n_buckets * 2); +} + +/* RCU callback used by ovs_flow_deferred_free. */ static void rcu_free_flow_callback(struct rcu_head *rcu) { struct sw_flow *flow = container_of(rcu, struct sw_flow, rcu); flow->dead = true; - flow_put(flow); + ovs_flow_put(flow); } /* Schedules 'flow' to be freed after the next RCU grace period. * The caller must hold rcu_read_lock for this to be sensible. */ -void flow_deferred_free(struct sw_flow *flow) +void ovs_flow_deferred_free(struct sw_flow *flow) { call_rcu(&flow->rcu, rcu_free_flow_callback); } -void flow_hold(struct sw_flow *flow) +void ovs_flow_hold(struct sw_flow *flow) { atomic_inc(&flow->refcnt); } -void flow_put(struct sw_flow *flow) +void ovs_flow_put(struct sw_flow *flow) { if (unlikely(!flow)) return; @@ -256,7 +442,7 @@ void flow_put(struct sw_flow *flow) } } -/* RCU callback used by flow_deferred_free_acts. */ +/* RCU callback used by ovs_flow_deferred_free_acts. */ static void rcu_free_acts_callback(struct rcu_head *rcu) { struct sw_flow_actions *sf_acts = container_of(rcu, @@ -266,7 +452,7 @@ static void rcu_free_acts_callback(struct rcu_head *rcu) /* Schedules 'sf_acts' to be freed after the next RCU grace period. * The caller must hold rcu_read_lock for this to be sensible. */ -void flow_deferred_free_acts(struct sw_flow_actions *sf_acts) +void ovs_flow_deferred_free_acts(struct sw_flow_actions *sf_acts) { call_rcu(&sf_acts->rcu, rcu_free_acts_callback); } @@ -279,6 +465,9 @@ static int parse_vlan(struct sk_buff *skb, struct sw_flow_key *key) }; struct qtag_prefix *qp; + if (unlikely(skb->len < sizeof(struct qtag_prefix) + sizeof(__be16))) + return 0; + if (unlikely(!pskb_may_pull(skb, sizeof(struct qtag_prefix) + sizeof(__be16)))) return -ENOMEM; @@ -358,13 +547,14 @@ static int parse_icmpv6(struct sk_buff *skb, struct sw_flow_key *key, } nd = (struct nd_msg *)skb_transport_header(skb); - ipv6_addr_copy(&key->ipv6.nd.target, &nd->target); + key->ipv6.nd.target = nd->target; key_len = SW_FLOW_KEY_OFFSET(ipv6.nd); icmp_len -= sizeof(*nd); offset = 0; while (icmp_len >= 8) { - struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)(nd->opt + offset); + struct nd_opt_hdr *nd_opt = + (struct nd_opt_hdr *)(nd->opt + offset); int opt_len = nd_opt->nd_opt_len * 8; if (unlikely(!opt_len || opt_len > icmp_len)) @@ -406,14 +596,12 @@ out: } /** - * flow_extract - extracts a flow key from an Ethernet frame. + * ovs_flow_extract - extracts a flow key from an Ethernet frame. * @skb: sk_buff that contains the frame, with skb->data pointing to the * Ethernet header * @in_port: port number on which @skb was received. * @key: output flow key * @key_lenp: length of output flow key - * @is_frag: set to 1 if @skb contains an IPv4 fragment, or to 0 if @skb does - * not contain an IPv4 packet or if it is not a fragment. * * The caller must ensure that skb->len >= ETH_HLEN. * @@ -431,17 +619,18 @@ out: * of a correct length, otherwise the same as skb->network_header. * For other key->dl_type values it is left untouched. */ -int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, - int *key_lenp, bool *is_frag) +int ovs_flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, + int *key_lenp) { int error = 0; int key_len = SW_FLOW_KEY_OFFSET(eth); struct ethhdr *eth; memset(key, 0, sizeof(*key)); - key->eth.tun_id = OVS_CB(skb)->tun_id; - key->eth.in_port = in_port; - *is_frag = false; + + key->phy.priority = skb->priority; + key->phy.tun_id = OVS_CB(skb)->tun_id; + key->phy.in_port = in_port; skb_reset_mac_header(skb); @@ -470,6 +659,7 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, /* Network layer. */ if (key->eth.type == htons(ETH_P_IP)) { struct iphdr *nh; + __be16 offset; key_len = SW_FLOW_KEY_OFFSET(ipv4.addr); @@ -485,35 +675,42 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, nh = ip_hdr(skb); key->ipv4.addr.src = nh->saddr; key->ipv4.addr.dst = nh->daddr; - key->ip.tos = nh->tos & ~INET_ECN_MASK; + key->ip.proto = nh->protocol; + key->ip.tos = nh->tos; + key->ip.ttl = nh->ttl; - /* Transport layer. */ - if ((nh->frag_off & htons(IP_MF | IP_OFFSET)) || - (skb_shinfo(skb)->gso_type & SKB_GSO_UDP)) - *is_frag = true; + offset = nh->frag_off & htons(IP_OFFSET); + if (offset) { + key->ip.frag = OVS_FRAG_TYPE_LATER; + goto out; + } + if (nh->frag_off & htons(IP_MF) || + skb_shinfo(skb)->gso_type & SKB_GSO_UDP) + key->ip.frag = OVS_FRAG_TYPE_FIRST; + /* Transport layer. */ if (key->ip.proto == IPPROTO_TCP) { key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (!*is_frag && tcphdr_ok(skb)) { + if (tcphdr_ok(skb)) { struct tcphdr *tcp = tcp_hdr(skb); key->ipv4.tp.src = tcp->source; key->ipv4.tp.dst = tcp->dest; } } else if (key->ip.proto == IPPROTO_UDP) { key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (!*is_frag && udphdr_ok(skb)) { + if (udphdr_ok(skb)) { struct udphdr *udp = udp_hdr(skb); key->ipv4.tp.src = udp->source; key->ipv4.tp.dst = udp->dest; } } else if (key->ip.proto == IPPROTO_ICMP) { key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (!*is_frag && icmphdr_ok(skb)) { + if (icmphdr_ok(skb)) { struct icmphdr *icmp = icmp_hdr(skb); /* The ICMP type and code fields use the 16-bit - * transport port fields, so we need to store them - * in 16-bit network byte order. */ + * transport port fields, so we need to store + * them in 16-bit network byte order. */ key->ipv4.tp.src = htons(icmp->type); key->ipv4.tp.dst = htons(icmp->code); } @@ -554,6 +751,11 @@ int flow_extract(struct sk_buff *skb, u16 in_port, struct sw_flow_key *key, goto out; } + if (key->ip.frag == OVS_FRAG_TYPE_LATER) + goto out; + if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP) + key->ip.frag = OVS_FRAG_TYPE_FIRST; + /* Transport layer. */ if (key->ip.proto == NEXTHDR_TCP) { key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); @@ -584,296 +786,375 @@ out: return error; } -u32 flow_hash(const struct sw_flow_key *key, int key_len) +u32 ovs_flow_hash(const struct sw_flow_key *key, int key_len) { - return jhash2((u32*)key, DIV_ROUND_UP(key_len, sizeof(u32)), hash_seed); + return jhash2((u32 *)key, DIV_ROUND_UP(key_len, sizeof(u32)), 0); +} + +struct sw_flow *ovs_flow_tbl_lookup(struct flow_table *table, + struct sw_flow_key *key, int key_len) +{ + struct sw_flow *flow; + struct hlist_node *n; + struct hlist_head *head; + u32 hash; + + hash = ovs_flow_hash(key, key_len); + + head = find_bucket(table, hash); + hlist_for_each_entry_rcu(flow, n, head, hash_node[table->node_ver]) { + + if (flow->hash == hash && + !memcmp(&flow->key, key, key_len)) { + return flow; + } + } + return NULL; } -int flow_cmp(const struct tbl_node *node, void *key2_, int len) +void ovs_flow_tbl_insert(struct flow_table *table, struct sw_flow *flow) { - const struct sw_flow_key *key1 = &flow_cast(node)->key; - const struct sw_flow_key *key2 = key2_; + struct hlist_head *head; + + head = find_bucket(table, flow->hash); + hlist_add_head_rcu(&flow->hash_node[table->node_ver], head); + table->count++; +} - return !memcmp(key1, key2, len); +void ovs_flow_tbl_remove(struct flow_table *table, struct sw_flow *flow) +{ + hlist_del_rcu(&flow->hash_node[table->node_ver]); + table->count--; + BUG_ON(table->count < 0); } -/* The size of the argument for each %ODP_KEY_ATTR_* Netlink attribute. */ -static const u32 key_lens[ODP_KEY_ATTR_MAX + 1] = { - [ODP_KEY_ATTR_TUN_ID] = 8, - [ODP_KEY_ATTR_IN_PORT] = 4, - [ODP_KEY_ATTR_ETHERNET] = sizeof(struct odp_key_ethernet), - [ODP_KEY_ATTR_8021Q] = sizeof(struct odp_key_8021q), - [ODP_KEY_ATTR_ETHERTYPE] = 2, - [ODP_KEY_ATTR_IPV4] = sizeof(struct odp_key_ipv4), - [ODP_KEY_ATTR_IPV6] = sizeof(struct odp_key_ipv6), - [ODP_KEY_ATTR_TCP] = sizeof(struct odp_key_tcp), - [ODP_KEY_ATTR_UDP] = sizeof(struct odp_key_udp), - [ODP_KEY_ATTR_ICMP] = sizeof(struct odp_key_icmp), - [ODP_KEY_ATTR_ICMPV6] = sizeof(struct odp_key_icmpv6), - [ODP_KEY_ATTR_ARP] = sizeof(struct odp_key_arp), - [ODP_KEY_ATTR_ND] = sizeof(struct odp_key_nd), +/* The size of the argument for each %OVS_KEY_ATTR_* Netlink attribute. */ +const int ovs_key_lens[OVS_KEY_ATTR_MAX + 1] = { + [OVS_KEY_ATTR_ENCAP] = -1, + [OVS_KEY_ATTR_PRIORITY] = sizeof(u32), + [OVS_KEY_ATTR_IN_PORT] = sizeof(u32), + [OVS_KEY_ATTR_ETHERNET] = sizeof(struct ovs_key_ethernet), + [OVS_KEY_ATTR_VLAN] = sizeof(__be16), + [OVS_KEY_ATTR_ETHERTYPE] = sizeof(__be16), + [OVS_KEY_ATTR_IPV4] = sizeof(struct ovs_key_ipv4), + [OVS_KEY_ATTR_IPV6] = sizeof(struct ovs_key_ipv6), + [OVS_KEY_ATTR_TCP] = sizeof(struct ovs_key_tcp), + [OVS_KEY_ATTR_UDP] = sizeof(struct ovs_key_udp), + [OVS_KEY_ATTR_ICMP] = sizeof(struct ovs_key_icmp), + [OVS_KEY_ATTR_ICMPV6] = sizeof(struct ovs_key_icmpv6), + [OVS_KEY_ATTR_ARP] = sizeof(struct ovs_key_arp), + [OVS_KEY_ATTR_ND] = sizeof(struct ovs_key_nd), + + /* Not upstream. */ + [OVS_KEY_ATTR_TUN_ID] = sizeof(__be64), }; +static int ipv4_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len, + const struct nlattr *a[], u64 *attrs) +{ + const struct ovs_key_icmp *icmp_key; + const struct ovs_key_tcp *tcp_key; + const struct ovs_key_udp *udp_key; + + switch (swkey->ip.proto) { + case IPPROTO_TCP: + if (!(*attrs & (1 << OVS_KEY_ATTR_TCP))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_TCP); + + *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); + tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]); + swkey->ipv4.tp.src = tcp_key->tcp_src; + swkey->ipv4.tp.dst = tcp_key->tcp_dst; + break; + + case IPPROTO_UDP: + if (!(*attrs & (1 << OVS_KEY_ATTR_UDP))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_UDP); + + *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); + udp_key = nla_data(a[OVS_KEY_ATTR_UDP]); + swkey->ipv4.tp.src = udp_key->udp_src; + swkey->ipv4.tp.dst = udp_key->udp_dst; + break; + + case IPPROTO_ICMP: + if (!(*attrs & (1 << OVS_KEY_ATTR_ICMP))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_ICMP); + + *key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); + icmp_key = nla_data(a[OVS_KEY_ATTR_ICMP]); + swkey->ipv4.tp.src = htons(icmp_key->icmp_type); + swkey->ipv4.tp.dst = htons(icmp_key->icmp_code); + break; + } + + return 0; +} + +static int ipv6_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_len, + const struct nlattr *a[], u64 *attrs) +{ + const struct ovs_key_icmpv6 *icmpv6_key; + const struct ovs_key_tcp *tcp_key; + const struct ovs_key_udp *udp_key; + + switch (swkey->ip.proto) { + case IPPROTO_TCP: + if (!(*attrs & (1 << OVS_KEY_ATTR_TCP))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_TCP); + + *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); + tcp_key = nla_data(a[OVS_KEY_ATTR_TCP]); + swkey->ipv6.tp.src = tcp_key->tcp_src; + swkey->ipv6.tp.dst = tcp_key->tcp_dst; + break; + + case IPPROTO_UDP: + if (!(*attrs & (1 << OVS_KEY_ATTR_UDP))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_UDP); + + *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); + udp_key = nla_data(a[OVS_KEY_ATTR_UDP]); + swkey->ipv6.tp.src = udp_key->udp_src; + swkey->ipv6.tp.dst = udp_key->udp_dst; + break; + + case IPPROTO_ICMPV6: + if (!(*attrs & (1 << OVS_KEY_ATTR_ICMPV6))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_ICMPV6); + + *key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); + icmpv6_key = nla_data(a[OVS_KEY_ATTR_ICMPV6]); + swkey->ipv6.tp.src = htons(icmpv6_key->icmpv6_type); + swkey->ipv6.tp.dst = htons(icmpv6_key->icmpv6_code); + + if (swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) || + swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) { + const struct ovs_key_nd *nd_key; + + if (!(*attrs & (1 << OVS_KEY_ATTR_ND))) + return -EINVAL; + *attrs &= ~(1 << OVS_KEY_ATTR_ND); + + *key_len = SW_FLOW_KEY_OFFSET(ipv6.nd); + nd_key = nla_data(a[OVS_KEY_ATTR_ND]); + memcpy(&swkey->ipv6.nd.target, nd_key->nd_target, + sizeof(swkey->ipv6.nd.target)); + memcpy(swkey->ipv6.nd.sll, nd_key->nd_sll, ETH_ALEN); + memcpy(swkey->ipv6.nd.tll, nd_key->nd_tll, ETH_ALEN); + } + break; + } + + return 0; +} + +static int parse_flow_nlattrs(const struct nlattr *attr, + const struct nlattr *a[], u64 *attrsp) +{ + const struct nlattr *nla; + u64 attrs; + int rem; + + attrs = 0; + nla_for_each_nested(nla, attr, rem) { + u16 type = nla_type(nla); + int expected_len; + + if (type > OVS_KEY_ATTR_MAX || attrs & (1ULL << type)) + return -EINVAL; + + expected_len = ovs_key_lens[type]; + if (nla_len(nla) != expected_len && expected_len != -1) + return -EINVAL; + + attrs |= 1ULL << type; + a[type] = nla; + } + if (rem) + return -EINVAL; + + *attrsp = attrs; + return 0; +} + /** - * flow_from_nlattrs - parses Netlink attributes into a flow key. + * ovs_flow_from_nlattrs - parses Netlink attributes into a flow key. * @swkey: receives the extracted flow key. * @key_lenp: number of bytes used in @swkey. - * @attr: Netlink attribute holding nested %ODP_KEY_ATTR_* Netlink attribute + * @attr: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute * sequence. - * - * This state machine accepts the following forms, with [] for optional - * elements and | for alternatives: - * - * [tun_id] in_port ethernet [8021q] [ethertype \ - * [IPv4 [TCP|UDP|ICMP] | IPv6 [TCP|UDP|ICMPv6 [ND]] | ARP]] */ -int flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp, +int ovs_flow_from_nlattrs(struct sw_flow_key *swkey, int *key_lenp, const struct nlattr *attr) { - int error = 0; - const struct nlattr *nla; - u16 prev_type; - int rem; + const struct nlattr *a[OVS_KEY_ATTR_MAX + 1]; + const struct ovs_key_ethernet *eth_key; int key_len; + u64 attrs; + int err; - memset(swkey, 0, sizeof(*swkey)); - swkey->eth.type = htons(ETH_P_802_2); + memset(swkey, 0, sizeof(struct sw_flow_key)); key_len = SW_FLOW_KEY_OFFSET(eth); - prev_type = ODP_KEY_ATTR_UNSPEC; - nla_for_each_nested(nla, attr, rem) { - const struct odp_key_ethernet *eth_key; - const struct odp_key_8021q *q_key; - const struct odp_key_ipv4 *ipv4_key; - const struct odp_key_ipv6 *ipv6_key; - const struct odp_key_tcp *tcp_key; - const struct odp_key_udp *udp_key; - const struct odp_key_icmp *icmp_key; - const struct odp_key_icmpv6 *icmpv6_key; - const struct odp_key_arp *arp_key; - const struct odp_key_nd *nd_key; - - int type = nla_type(nla); - - if (type > ODP_KEY_ATTR_MAX || nla_len(nla) != key_lens[type]) - goto invalid; - -#define TRANSITION(PREV_TYPE, TYPE) (((PREV_TYPE) << 16) | (TYPE)) - switch (TRANSITION(prev_type, type)) { - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_TUN_ID): - swkey->eth.tun_id = nla_get_be64(nla); - break; - - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_IN_PORT): - case TRANSITION(ODP_KEY_ATTR_TUN_ID, ODP_KEY_ATTR_IN_PORT): - if (nla_get_u32(nla) >= DP_MAX_PORTS) - goto invalid; - swkey->eth.in_port = nla_get_u32(nla); - break; - - case TRANSITION(ODP_KEY_ATTR_IN_PORT, ODP_KEY_ATTR_ETHERNET): - eth_key = nla_data(nla); - memcpy(swkey->eth.src, eth_key->eth_src, ETH_ALEN); - memcpy(swkey->eth.dst, eth_key->eth_dst, ETH_ALEN); - break; - - case TRANSITION(ODP_KEY_ATTR_ETHERNET, ODP_KEY_ATTR_8021Q): - q_key = nla_data(nla); - /* Only standard 0x8100 VLANs currently supported. */ - if (q_key->q_tpid != htons(ETH_P_8021Q)) - goto invalid; - if (q_key->q_tci & htons(VLAN_TAG_PRESENT)) - goto invalid; - swkey->eth.tci = q_key->q_tci | htons(VLAN_TAG_PRESENT); - break; + err = parse_flow_nlattrs(attr, a, &attrs); + if (err) + return err; - case TRANSITION(ODP_KEY_ATTR_8021Q, ODP_KEY_ATTR_ETHERTYPE): - case TRANSITION(ODP_KEY_ATTR_ETHERNET, ODP_KEY_ATTR_ETHERTYPE): - swkey->eth.type = nla_get_be16(nla); - if (ntohs(swkey->eth.type) < 1536) - goto invalid; - break; + /* Metadata attributes. */ + if (attrs & (1 << OVS_KEY_ATTR_PRIORITY)) { + swkey->phy.priority = nla_get_u32(a[OVS_KEY_ATTR_PRIORITY]); + attrs &= ~(1 << OVS_KEY_ATTR_PRIORITY); + } + if (attrs & (1 << OVS_KEY_ATTR_IN_PORT)) { + u32 in_port = nla_get_u32(a[OVS_KEY_ATTR_IN_PORT]); + if (in_port >= DP_MAX_PORTS) + return -EINVAL; + swkey->phy.in_port = in_port; + attrs &= ~(1 << OVS_KEY_ATTR_IN_PORT); + } else { + swkey->phy.in_port = DP_MAX_PORTS; + } - case TRANSITION(ODP_KEY_ATTR_ETHERTYPE, ODP_KEY_ATTR_IPV4): - key_len = SW_FLOW_KEY_OFFSET(ipv4.addr); - if (swkey->eth.type != htons(ETH_P_IP)) - goto invalid; - ipv4_key = nla_data(nla); - swkey->ip.proto = ipv4_key->ipv4_proto; - swkey->ip.tos = ipv4_key->ipv4_tos; - swkey->ipv4.addr.src = ipv4_key->ipv4_src; - swkey->ipv4.addr.dst = ipv4_key->ipv4_dst; - if (swkey->ip.tos & INET_ECN_MASK) - goto invalid; - break; + if (attrs & (1ULL << OVS_KEY_ATTR_TUN_ID)) { + swkey->phy.tun_id = nla_get_be64(a[OVS_KEY_ATTR_TUN_ID]); + attrs &= ~(1ULL << OVS_KEY_ATTR_TUN_ID); + } - case TRANSITION(ODP_KEY_ATTR_ETHERTYPE, ODP_KEY_ATTR_IPV6): - key_len = SW_FLOW_KEY_OFFSET(ipv6.addr); - if (swkey->eth.type != htons(ETH_P_IPV6)) - goto invalid; - ipv6_key = nla_data(nla); - swkey->ip.proto = ipv6_key->ipv6_proto; - swkey->ip.tos = ipv6_key->ipv6_tos; - memcpy(&swkey->ipv6.addr.src, ipv6_key->ipv6_src, - sizeof(swkey->ipv6.addr.src)); - memcpy(&swkey->ipv6.addr.dst, ipv6_key->ipv6_dst, - sizeof(swkey->ipv6.addr.dst)); - if (swkey->ip.tos & INET_ECN_MASK) - goto invalid; - break; + /* Data attributes. */ + if (!(attrs & (1 << OVS_KEY_ATTR_ETHERNET))) + return -EINVAL; + attrs &= ~(1 << OVS_KEY_ATTR_ETHERNET); - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_TCP): - key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (swkey->ip.proto != IPPROTO_TCP) - goto invalid; - tcp_key = nla_data(nla); - swkey->ipv4.tp.src = tcp_key->tcp_src; - swkey->ipv4.tp.dst = tcp_key->tcp_dst; - break; + eth_key = nla_data(a[OVS_KEY_ATTR_ETHERNET]); + memcpy(swkey->eth.src, eth_key->eth_src, ETH_ALEN); + memcpy(swkey->eth.dst, eth_key->eth_dst, ETH_ALEN); - case TRANSITION(ODP_KEY_ATTR_IPV6, ODP_KEY_ATTR_TCP): - key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); - if (swkey->ip.proto != IPPROTO_TCP) - goto invalid; - tcp_key = nla_data(nla); - swkey->ipv6.tp.src = tcp_key->tcp_src; - swkey->ipv6.tp.dst = tcp_key->tcp_dst; - break; + if (attrs & (1u << OVS_KEY_ATTR_ETHERTYPE) && + nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]) == htons(ETH_P_8021Q)) { + const struct nlattr *encap; + __be16 tci; - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_UDP): - key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (swkey->ip.proto != IPPROTO_UDP) - goto invalid; - udp_key = nla_data(nla); - swkey->ipv4.tp.src = udp_key->udp_src; - swkey->ipv4.tp.dst = udp_key->udp_dst; - break; + if (attrs != ((1 << OVS_KEY_ATTR_VLAN) | + (1 << OVS_KEY_ATTR_ETHERTYPE) | + (1 << OVS_KEY_ATTR_ENCAP))) + return -EINVAL; + + encap = a[OVS_KEY_ATTR_ENCAP]; + tci = nla_get_be16(a[OVS_KEY_ATTR_VLAN]); + if (tci & htons(VLAN_TAG_PRESENT)) { + swkey->eth.tci = tci; + + err = parse_flow_nlattrs(encap, a, &attrs); + if (err) + return err; + } else if (!tci) { + /* Corner case for truncated 802.1Q header. */ + if (nla_len(encap)) + return -EINVAL; - case TRANSITION(ODP_KEY_ATTR_IPV6, ODP_KEY_ATTR_UDP): - key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); - if (swkey->ip.proto != IPPROTO_UDP) - goto invalid; - udp_key = nla_data(nla); - swkey->ipv6.tp.src = udp_key->udp_src; - swkey->ipv6.tp.dst = udp_key->udp_dst; - break; + swkey->eth.type = htons(ETH_P_8021Q); + *key_lenp = key_len; + return 0; + } else { + return -EINVAL; + } + } - case TRANSITION(ODP_KEY_ATTR_IPV4, ODP_KEY_ATTR_ICMP): - key_len = SW_FLOW_KEY_OFFSET(ipv4.tp); - if (swkey->ip.proto != IPPROTO_ICMP) - goto invalid; - icmp_key = nla_data(nla); - swkey->ipv4.tp.src = htons(icmp_key->icmp_type); - swkey->ipv4.tp.dst = htons(icmp_key->icmp_code); - break; + if (attrs & (1 << OVS_KEY_ATTR_ETHERTYPE)) { + swkey->eth.type = nla_get_be16(a[OVS_KEY_ATTR_ETHERTYPE]); + if (ntohs(swkey->eth.type) < 1536) + return -EINVAL; + attrs &= ~(1 << OVS_KEY_ATTR_ETHERTYPE); + } else { + swkey->eth.type = htons(ETH_P_802_2); + } - case TRANSITION(ODP_KEY_ATTR_IPV6, ODP_KEY_ATTR_ICMPV6): - key_len = SW_FLOW_KEY_OFFSET(ipv6.tp); - if (swkey->ip.proto != IPPROTO_ICMPV6) - goto invalid; - icmpv6_key = nla_data(nla); - swkey->ipv6.tp.src = htons(icmpv6_key->icmpv6_type); - swkey->ipv6.tp.dst = htons(icmpv6_key->icmpv6_code); - break; + if (swkey->eth.type == htons(ETH_P_IP)) { + const struct ovs_key_ipv4 *ipv4_key; - case TRANSITION(ODP_KEY_ATTR_ETHERTYPE, ODP_KEY_ATTR_ARP): - key_len = SW_FLOW_KEY_OFFSET(ipv4.arp); - if (swkey->eth.type != htons(ETH_P_ARP)) - goto invalid; - arp_key = nla_data(nla); - swkey->ipv4.addr.src = arp_key->arp_sip; - swkey->ipv4.addr.dst = arp_key->arp_tip; - if (arp_key->arp_op & htons(0xff00)) - goto invalid; - swkey->ip.proto = ntohs(arp_key->arp_op); - memcpy(swkey->ipv4.arp.sha, arp_key->arp_sha, ETH_ALEN); - memcpy(swkey->ipv4.arp.tha, arp_key->arp_tha, ETH_ALEN); - break; - - case TRANSITION(ODP_KEY_ATTR_ICMPV6, ODP_KEY_ATTR_ND): - key_len = SW_FLOW_KEY_OFFSET(ipv6.nd); - if (swkey->ipv6.tp.src != htons(NDISC_NEIGHBOUR_SOLICITATION) - && swkey->ipv6.tp.src != htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) - goto invalid; - nd_key = nla_data(nla); - memcpy(&swkey->ipv6.nd.target, nd_key->nd_target, - sizeof(swkey->ipv6.nd.target)); - memcpy(swkey->ipv6.nd.sll, nd_key->nd_sll, ETH_ALEN); - memcpy(swkey->ipv6.nd.tll, nd_key->nd_tll, ETH_ALEN); - break; + if (!(attrs & (1 << OVS_KEY_ATTR_IPV4))) + return -EINVAL; + attrs &= ~(1 << OVS_KEY_ATTR_IPV4); - default: - goto invalid; + key_len = SW_FLOW_KEY_OFFSET(ipv4.addr); + ipv4_key = nla_data(a[OVS_KEY_ATTR_IPV4]); + if (ipv4_key->ipv4_frag > OVS_FRAG_TYPE_MAX) + return -EINVAL; + swkey->ip.proto = ipv4_key->ipv4_proto; + swkey->ip.tos = ipv4_key->ipv4_tos; + swkey->ip.ttl = ipv4_key->ipv4_ttl; + swkey->ip.frag = ipv4_key->ipv4_frag; + swkey->ipv4.addr.src = ipv4_key->ipv4_src; + swkey->ipv4.addr.dst = ipv4_key->ipv4_dst; + + if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) { + err = ipv4_flow_from_nlattrs(swkey, &key_len, a, &attrs); + if (err) + return err; } - - prev_type = type; - } - if (rem) - goto invalid; - - switch (prev_type) { - case ODP_KEY_ATTR_UNSPEC: - goto invalid; - - case ODP_KEY_ATTR_TUN_ID: - case ODP_KEY_ATTR_IN_PORT: - goto invalid; - - case ODP_KEY_ATTR_ETHERNET: - case ODP_KEY_ATTR_8021Q: - goto ok; - - case ODP_KEY_ATTR_ETHERTYPE: - if (swkey->eth.type == htons(ETH_P_IP) || - swkey->eth.type == htons(ETH_P_ARP)) - goto invalid; - goto ok; - - case ODP_KEY_ATTR_IPV4: - if (swkey->ip.proto == IPPROTO_TCP || - swkey->ip.proto == IPPROTO_UDP || - swkey->ip.proto == IPPROTO_ICMP) - goto invalid; - goto ok; - - case ODP_KEY_ATTR_IPV6: - if (swkey->ip.proto == IPPROTO_TCP || - swkey->ip.proto == IPPROTO_UDP || - swkey->ip.proto == IPPROTO_ICMPV6) - goto invalid; - goto ok; - - case ODP_KEY_ATTR_ICMPV6: - if (swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_SOLICITATION) || - swkey->ipv6.tp.src == htons(NDISC_NEIGHBOUR_ADVERTISEMENT)) - goto invalid; - goto ok; - - case ODP_KEY_ATTR_TCP: - case ODP_KEY_ATTR_UDP: - case ODP_KEY_ATTR_ICMP: - case ODP_KEY_ATTR_ARP: - case ODP_KEY_ATTR_ND: - goto ok; - - default: - WARN_ON_ONCE(1); + } else if (swkey->eth.type == htons(ETH_P_IPV6)) { + const struct ovs_key_ipv6 *ipv6_key; + + if (!(attrs & (1 << OVS_KEY_ATTR_IPV6))) + return -EINVAL; + attrs &= ~(1 << OVS_KEY_ATTR_IPV6); + + key_len = SW_FLOW_KEY_OFFSET(ipv6.label); + ipv6_key = nla_data(a[OVS_KEY_ATTR_IPV6]); + if (ipv6_key->ipv6_frag > OVS_FRAG_TYPE_MAX) + return -EINVAL; + swkey->ipv6.label = ipv6_key->ipv6_label; + swkey->ip.proto = ipv6_key->ipv6_proto; + swkey->ip.tos = ipv6_key->ipv6_tclass; + swkey->ip.ttl = ipv6_key->ipv6_hlimit; + swkey->ip.frag = ipv6_key->ipv6_frag; + memcpy(&swkey->ipv6.addr.src, ipv6_key->ipv6_src, + sizeof(swkey->ipv6.addr.src)); + memcpy(&swkey->ipv6.addr.dst, ipv6_key->ipv6_dst, + sizeof(swkey->ipv6.addr.dst)); + + if (swkey->ip.frag != OVS_FRAG_TYPE_LATER) { + err = ipv6_flow_from_nlattrs(swkey, &key_len, a, &attrs); + if (err) + return err; + } + } else if (swkey->eth.type == htons(ETH_P_ARP)) { + const struct ovs_key_arp *arp_key; + + if (!(attrs & (1 << OVS_KEY_ATTR_ARP))) + return -EINVAL; + attrs &= ~(1 << OVS_KEY_ATTR_ARP); + + key_len = SW_FLOW_KEY_OFFSET(ipv4.arp); + arp_key = nla_data(a[OVS_KEY_ATTR_ARP]); + swkey->ipv4.addr.src = arp_key->arp_sip; + swkey->ipv4.addr.dst = arp_key->arp_tip; + if (arp_key->arp_op & htons(0xff00)) + return -EINVAL; + swkey->ip.proto = ntohs(arp_key->arp_op); + memcpy(swkey->ipv4.arp.sha, arp_key->arp_sha, ETH_ALEN); + memcpy(swkey->ipv4.arp.tha, arp_key->arp_tha, ETH_ALEN); } -invalid: - error = -EINVAL; - -ok: - WARN_ON_ONCE(!key_len && !error); + if (attrs) + return -EINVAL; *key_lenp = key_len; - return error; + + return 0; } /** - * flow_metadata_from_nlattrs - parses Netlink attributes into a flow key. + * ovs_flow_metadata_from_nlattrs - parses Netlink attributes into a flow key. * @in_port: receives the extracted input port. * @tun_id: receives the extracted tunnel ID. - * @key: Netlink attribute holding nested %ODP_KEY_ATTR_* Netlink attribute + * @key: Netlink attribute holding nested %OVS_KEY_ATTR_* Netlink attribute * sequence. * * This parses a series of Netlink attributes that form a flow key, which must @@ -881,119 +1162,123 @@ ok: * get the metadata, that is, the parts of the flow key that cannot be * extracted from the packet itself. */ -int flow_metadata_from_nlattrs(u16 *in_port, __be64 *tun_id, - const struct nlattr *attr) +int ovs_flow_metadata_from_nlattrs(u32 *priority, u16 *in_port, __be64 *tun_id, + const struct nlattr *attr) { const struct nlattr *nla; - u16 prev_type; int rem; + *in_port = DP_MAX_PORTS; *tun_id = 0; + *priority = 0; - prev_type = ODP_KEY_ATTR_UNSPEC; nla_for_each_nested(nla, attr, rem) { - int type = nla_type(nla); + int type = nla_type(nla); - if (type > ODP_KEY_ATTR_MAX || nla_len(nla) != key_lens[type]) - return -EINVAL; + if (type <= OVS_KEY_ATTR_MAX && ovs_key_lens[type] > 0) { + if (nla_len(nla) != ovs_key_lens[type]) + return -EINVAL; - switch (TRANSITION(prev_type, type)) { - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_TUN_ID): - *tun_id = nla_get_be64(nla); - break; + switch (type) { + case OVS_KEY_ATTR_PRIORITY: + *priority = nla_get_u32(nla); + break; - case TRANSITION(ODP_KEY_ATTR_UNSPEC, ODP_KEY_ATTR_IN_PORT): - case TRANSITION(ODP_KEY_ATTR_TUN_ID, ODP_KEY_ATTR_IN_PORT): - if (nla_get_u32(nla) >= DP_MAX_PORTS) - return -EINVAL; - *in_port = nla_get_u32(nla); - break; + case OVS_KEY_ATTR_TUN_ID: + *tun_id = nla_get_be64(nla); + break; - default: - goto done; + case OVS_KEY_ATTR_IN_PORT: + if (nla_get_u32(nla) >= DP_MAX_PORTS) + return -EINVAL; + *in_port = nla_get_u32(nla); + break; + } } - - prev_type = type; } if (rem) return -EINVAL; - -done: - if (prev_type == ODP_KEY_ATTR_UNSPEC || - prev_type == ODP_KEY_ATTR_TUN_ID) - return -EINVAL; return 0; } -int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) +int ovs_flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) { - struct odp_key_ethernet *eth_key; - struct nlattr *nla; + struct ovs_key_ethernet *eth_key; + struct nlattr *nla, *encap; - /* This is an imperfect sanity-check that FLOW_BUFSIZE doesn't need - * to be updated, but will at least raise awareness when new ODP key - * types are added. */ - BUILD_BUG_ON(__ODP_KEY_ATTR_MAX != 14); + if (swkey->phy.priority && + nla_put_u32(skb, OVS_KEY_ATTR_PRIORITY, swkey->phy.priority)) + goto nla_put_failure; - if (swkey->eth.tun_id != cpu_to_be64(0)) - NLA_PUT_BE64(skb, ODP_KEY_ATTR_TUN_ID, swkey->eth.tun_id); + if (swkey->phy.tun_id != cpu_to_be64(0) && + nla_put_be64(skb, OVS_KEY_ATTR_TUN_ID, swkey->phy.tun_id)) + goto nla_put_failure; - NLA_PUT_U32(skb, ODP_KEY_ATTR_IN_PORT, swkey->eth.in_port); + if (swkey->phy.in_port != DP_MAX_PORTS && + nla_put_u32(skb, OVS_KEY_ATTR_IN_PORT, swkey->phy.in_port)) + goto nla_put_failure; - nla = nla_reserve(skb, ODP_KEY_ATTR_ETHERNET, sizeof(*eth_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_ETHERNET, sizeof(*eth_key)); if (!nla) goto nla_put_failure; eth_key = nla_data(nla); memcpy(eth_key->eth_src, swkey->eth.src, ETH_ALEN); memcpy(eth_key->eth_dst, swkey->eth.dst, ETH_ALEN); - if (swkey->eth.tci != htons(0)) { - struct odp_key_8021q q_key; - - q_key.q_tpid = htons(ETH_P_8021Q); - q_key.q_tci = swkey->eth.tci & ~htons(VLAN_TAG_PRESENT); - NLA_PUT(skb, ODP_KEY_ATTR_8021Q, sizeof(q_key), &q_key); + if (swkey->eth.tci || swkey->eth.type == htons(ETH_P_8021Q)) { + if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, htons(ETH_P_8021Q)) || + nla_put_be16(skb, OVS_KEY_ATTR_VLAN, swkey->eth.tci)) + goto nla_put_failure; + encap = nla_nest_start(skb, OVS_KEY_ATTR_ENCAP); + if (!swkey->eth.tci) + goto unencap; + } else { + encap = NULL; } if (swkey->eth.type == htons(ETH_P_802_2)) - return 0; + goto unencap; - NLA_PUT_BE16(skb, ODP_KEY_ATTR_ETHERTYPE, swkey->eth.type); + if (nla_put_be16(skb, OVS_KEY_ATTR_ETHERTYPE, swkey->eth.type)) + goto nla_put_failure; if (swkey->eth.type == htons(ETH_P_IP)) { - struct odp_key_ipv4 *ipv4_key; + struct ovs_key_ipv4 *ipv4_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_IPV4, sizeof(*ipv4_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_IPV4, sizeof(*ipv4_key)); if (!nla) goto nla_put_failure; ipv4_key = nla_data(nla); - memset(ipv4_key, 0, sizeof(struct odp_key_ipv4)); ipv4_key->ipv4_src = swkey->ipv4.addr.src; ipv4_key->ipv4_dst = swkey->ipv4.addr.dst; ipv4_key->ipv4_proto = swkey->ip.proto; ipv4_key->ipv4_tos = swkey->ip.tos; + ipv4_key->ipv4_ttl = swkey->ip.ttl; + ipv4_key->ipv4_frag = swkey->ip.frag; } else if (swkey->eth.type == htons(ETH_P_IPV6)) { - struct odp_key_ipv6 *ipv6_key; + struct ovs_key_ipv6 *ipv6_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_IPV6, sizeof(*ipv6_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_IPV6, sizeof(*ipv6_key)); if (!nla) goto nla_put_failure; ipv6_key = nla_data(nla); - memset(ipv6_key, 0, sizeof(struct odp_key_ipv6)); memcpy(ipv6_key->ipv6_src, &swkey->ipv6.addr.src, sizeof(ipv6_key->ipv6_src)); memcpy(ipv6_key->ipv6_dst, &swkey->ipv6.addr.dst, sizeof(ipv6_key->ipv6_dst)); + ipv6_key->ipv6_label = swkey->ipv6.label; ipv6_key->ipv6_proto = swkey->ip.proto; - ipv6_key->ipv6_tos = swkey->ip.tos; + ipv6_key->ipv6_tclass = swkey->ip.tos; + ipv6_key->ipv6_hlimit = swkey->ip.ttl; + ipv6_key->ipv6_frag = swkey->ip.frag; } else if (swkey->eth.type == htons(ETH_P_ARP)) { - struct odp_key_arp *arp_key; + struct ovs_key_arp *arp_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_ARP, sizeof(*arp_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_ARP, sizeof(*arp_key)); if (!nla) goto nla_put_failure; arp_key = nla_data(nla); - memset(arp_key, 0, sizeof(struct odp_key_arp)); + memset(arp_key, 0, sizeof(struct ovs_key_arp)); arp_key->arp_sip = swkey->ipv4.addr.src; arp_key->arp_tip = swkey->ipv4.addr.dst; arp_key->arp_op = htons(swkey->ip.proto); @@ -1001,13 +1286,14 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) memcpy(arp_key->arp_tha, swkey->ipv4.arp.tha, ETH_ALEN); } - if (swkey->eth.type == htons(ETH_P_IP) || - swkey->eth.type == htons(ETH_P_IPV6)) { + if ((swkey->eth.type == htons(ETH_P_IP) || + swkey->eth.type == htons(ETH_P_IPV6)) && + swkey->ip.frag != OVS_FRAG_TYPE_LATER) { if (swkey->ip.proto == IPPROTO_TCP) { - struct odp_key_tcp *tcp_key; + struct ovs_key_tcp *tcp_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_TCP, sizeof(*tcp_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_TCP, sizeof(*tcp_key)); if (!nla) goto nla_put_failure; tcp_key = nla_data(nla); @@ -1019,9 +1305,9 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) tcp_key->tcp_dst = swkey->ipv6.tp.dst; } } else if (swkey->ip.proto == IPPROTO_UDP) { - struct odp_key_udp *udp_key; + struct ovs_key_udp *udp_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_UDP, sizeof(*udp_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_UDP, sizeof(*udp_key)); if (!nla) goto nla_put_failure; udp_key = nla_data(nla); @@ -1034,9 +1320,9 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) } } else if (swkey->eth.type == htons(ETH_P_IP) && swkey->ip.proto == IPPROTO_ICMP) { - struct odp_key_icmp *icmp_key; + struct ovs_key_icmp *icmp_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_ICMP, sizeof(*icmp_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_ICMP, sizeof(*icmp_key)); if (!nla) goto nla_put_failure; icmp_key = nla_data(nla); @@ -1044,9 +1330,9 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) icmp_key->icmp_code = ntohs(swkey->ipv4.tp.dst); } else if (swkey->eth.type == htons(ETH_P_IPV6) && swkey->ip.proto == IPPROTO_ICMPV6) { - struct odp_key_icmpv6 *icmpv6_key; + struct ovs_key_icmpv6 *icmpv6_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_ICMPV6, + nla = nla_reserve(skb, OVS_KEY_ATTR_ICMPV6, sizeof(*icmpv6_key)); if (!nla) goto nla_put_failure; @@ -1056,9 +1342,9 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) if (icmpv6_key->icmpv6_type == NDISC_NEIGHBOUR_SOLICITATION || icmpv6_key->icmpv6_type == NDISC_NEIGHBOUR_ADVERTISEMENT) { - struct odp_key_nd *nd_key; + struct ovs_key_nd *nd_key; - nla = nla_reserve(skb, ODP_KEY_ATTR_ND, sizeof(*nd_key)); + nla = nla_reserve(skb, OVS_KEY_ATTR_ND, sizeof(*nd_key)); if (!nla) goto nla_put_failure; nd_key = nla_data(nla); @@ -1070,6 +1356,10 @@ int flow_to_nlattrs(const struct sw_flow_key *swkey, struct sk_buff *skb) } } +unencap: + if (encap) + nla_nest_end(skb, encap); + return 0; nla_put_failure: @@ -1078,20 +1368,18 @@ nla_put_failure: /* Initializes the flow module. * Returns zero if successful or a negative error code. */ -int flow_init(void) +int ovs_flow_init(void) { flow_cache = kmem_cache_create("sw_flow", sizeof(struct sw_flow), 0, 0, NULL); if (flow_cache == NULL) return -ENOMEM; - get_random_bytes(&hash_seed, sizeof(hash_seed)); - return 0; } /* Uninitializes the flow module. */ -void flow_exit(void) +void ovs_flow_exit(void) { kmem_cache_destroy(flow_cache); }